Estimating rates of movement among populations is never simple, and where young animals cannot all be captured at their birth sites, traditional field methods potentially underestimate dispersal rates. Genetic assignment tests appear to hold promise for detecting ‘precapture’ dispersal, and recent evidence suggests that even on the scale of dispersal between populations, genetic parentage analyses can also be informative. Herein, we examine the performance of both types of analysis with data from a 17-year study of dispersal in banner-tailed kangaroo rats Dipodomys spectabilis. We compare estimates of precapture dispersal from (i) the commonly used parentage analysis program cervus (ii) a pedigree-reconstruction program, MasterBayes, that combines genetic with spatial and other nongenetic information and (iii) genetic assignment procedures implemented by the program geneclass2, with (iv) rates of dispersal observed through recapture of a subset of animals initially marked shortly after weaning. geneclass2 estimates a larger proportion of precapture dispersers than MasterBayes, but both approaches as well as those based on field data alone, suggest that approximately 10% of adults in local populations are immigrants and that interpopulation dispersal is slightly female-biassed. All genetic procedures detect precapture dispersal between populations, but dispersers identified by MasterBayes are particularly compatible with what is independently known about body mass at dispersal, dispersal distance and distance between parents. Parentage analyses have considerable potential to infer the value of this otherwise elusive demographic parameter when most candidate parents can be genotyped and when nongenetic information, especially the distance separating candidate mothers and fathers, can be incorporated into the procedure.